SleekAI Changelog

document inline chatbot shortcode aliases, public permalink templates, retrieval priorities, and REST create and replacement behavior

authenticate administrative completion, model, streaming test, and sync routes while preserving signed public chatbot requests

automatically create missing vector stores from existing synchronized chatbot files on upgrade, with retries and a first-request repair instead of waiting for the next scheduled sync

cache-bust public widget scripts after plugin updates so browsers load the corrected response parsers

expose registered custom chatbot conditions in the editor, support legacy boolean values, and fail closed for unknown conditions

finish Gemini replies in polling mode, preserve response events across partial reads, and display all returned text parts

include the configured provider and model in public chatbot widgets so Google and Anthropic replies use the correct parser

keep chatbot title generation separate from knowledge-base retrieval, main instructions, and conversation logs

log user messages and completed assistant replies for synchronized file-search chatbots in both streaming and polling mode, including older widget chat IDs

mount type=”embed” shortcodes as inline chats, including initially hidden hosts, without duplicate React roots

preserve current page, user, taxonomy, and term data in public chatbot requests instead of using admin-preview selections

preserve file-search reply events split across network chunks and surface provider failures without logging them as successful replies

preserve full model names and support both legacy and current model IDs in chatbot previews and widgets

schedule recurring knowledge syncs after the selected interval instead of duplicating the initial inline upload

use OpenAI file search without also injecting synchronized query data inline, and preserve the previous file store if a refresh fails

wait for successful file indexing before replacing a chatbot vector store, preserving the previous store if indexing fails

opt-in query and content-type retrieval priorities with per-post metadata, bounded reranking, and developer weight overrides

public permalink fields in query templates and URL-aware defaults for new Wizard data

prevent anonymous callers from selecting arbitrary models or consuming configured API keys through the completions endpoint; thanks to Serwis4U for the responsible disclosure

detect installed coding-agent providers and show explicit installation and local bridge connection errors

keep coding-agent workspaces durable and resume Claude Code or Codex sessions after local agent restarts

keep scheduled retention callbacks and workflow signatures compatible with PHP 7.4

validate shared admin settings payloads and sanitize license and nonce inputs consistently

Codex terminal support alongside Claude Code in page-builder agent modes

Etch coding-agent integration with builder-specific context and runtime tools

scheduled chat retention with manual cleanup through `wp sleek ai cleanup retention`

WP-CLI commands under wp sleek ai for API keys, settings, alt text, chatbots, agents and models

normalize backend boot, providers, lifecycle ownership, and standalone isolation for SleekOS module compatibility

OpenAI-compatible custom model endpoints

scope Tailwind preflight styles to the plugin UI to avoid affecting WordPress pages

terminal conversations in page-builder agent modes now share the same builder files, so users can continue work across multiple chats and terminals

Agent modes for Bricks, Elementor, Breakdance, and Beaver Builder

Claude Code terminal support inside page-builder agent modes

file-based post type sync for chatbots, agents, and agent modes via `sleek/ai` in the active theme

generated `AGENTS.md` files in synced theme folders from plugin-side instructions and schemas

page-builder agents can inspect the current page, edit existing content and styles, add new sections/elements, move/duplicate/delete elements, and save builder changes

page-builder agents can work with builder-specific design features where available, including custom CSS, classes/selectors, variables/tokens, global styles, templates/components, responsive settings, dynamic data, and media controls

use new logo

minor chatbot bugs

file casing

API keys need to be re-entered after upgrade

Everywhere mode (use Agent instead)

Plugin mode standalone (use Agent instead)

use cache busting for API endpoints

Agent

Chat – branch conversations from any message

Chat – media library integration for file context

Chat – regenerate messages

Chatbot – automatic file regeneration on schedule

Chatbot – file-based context system (replaces Assistant mode)

Chatbot – IndexedDB storage for persistent history

Chatbot – widget rebuilt on new chat foundation

Chatbot – Wizard mode for easy setup

complete rebuild using unified chat framework

dedicated database table for chats (improved performance)

chat persistance not working

error messages from provider request not passed down to UI

widget chat button cropped if distance bigger than 16px

streaming potentially not working on dev sites

better autoscroll UX

activate license via constant

whitelabel functionality

undefined showing in parameter label description

alt text generation not respecting system instruction

add o4 to model dropdown

filter out non compatible models from plugin model dropdown

chatbot logging not correctly saving in a few instances

‘sleekAi_generate_response’ function to programtically generate LLM responses

‘sleekAi/settings’ filter

‘sleekAi/settings/user’ filter

image alt tag generation

back button going back to home if in a nested route

Everywhere mode not working on TinyMCE inside iframs

WordPress 6.8 compatibility

chat input firing on Enter when using IMEs

Everywhere mode not working when using TinyMCE

Chatbot logs showing in admin area

error about missing system instructions when using Assistants

settings in chatbot not center aligned

ability to save chatbot logs

use Anthropic endpoint for model selection

Google Gemini Support

OpenRouter support

automatically separate items in file creation for OpenAI Assistants with three dashes (—)

show error boundary if something goes wrong

show shortcode with embed type on chatbot page

sidebar items are now links

unlimited number of chatbots can be rendered at the same time

instruction to dynamic data variables

return stringified JSON if using an object in data source

‘userMeta’ data source retrieving user data instead of user meta

don’t render plugins data in script response

incompatibility with PHP 7.4

terms query not working correctly in terms

add ACF data to data source if available

add MB data to data source if available

list rendering in chats

list rendering in chats

logo alignment in chatbot

don’t show plugins CPT

plugin not deploying correctly

back button will always go back to base

don’t show audio and realtime models

set default OpenAI model to gpt-4o

sticky code block headers

chatbot not locking body anymore

titles not being generated when using Anthropic

plugin generation

support for o3 models

dev modules loaded in admin

autoscroll misbehaving

version fix

version fix

everywhere not working when site editor inside iframe

custom conditions not working as booleans

add latest assets

allow scrolling up when completion is ongoing with streaming enabled

confirm deletion before deleting chat or chatbot

lock submit button to top right of input

lots of UI improvements

more performant auto-scroller

o1 models not working

scrollbar not showing when sending long message

Tables not rendering correctly in chats

add custom conditions to chatbots

add custom sources to chatbots

Anthropic API support

possibility to persist chatbot messages between page navigations (sessionStorage)

make chats wider

chatbot options not working when embedding chatbot

lists in messages

add home icon to sidebar logo

chatbot UI fixes

Assistants v2 API not working when no vector store is set

apply ‘text-balance -> pretty’ to message text elements

show o1 models in model dropdown

use Assistants v2 API

new message reloading once when using streaming

combine SleekWP products in admin sidebar

not equal operator not working in conditions

some UI fixes

19 chatbot conditions

better Skeleton UI

don’t remove send button on mobile

don’t show fullscreen button on mobile

rotate button arrow to right

update store url

chatbot input size potentially growing too big with no value set

SleekAI.chatbot.hide() function

SleekAI.chatbot.show() function

current post not used as ID when set in Chatbot data source

custom WP_Query not working as Chatbot data source

parameter combobox not working properly for Chatbot conditions

autoheight textboxes jumping back to top

chatbot presets not wrapping

chatbot input height being miscalculated

modernise textbox UI

modernise widget UI

.arrow CSS styles leaking into code fields when in WordPress admin

links with hashes would open widget if triggered before

‘alignment’ setting for widget

‘border-radius’ setting for widget

‘distance’ setting for widget

‘fullscreen’ setting for widget

Everywhere – custom command

Everywhere – presets

license showing after being deleted in edge cases

UI fixes

don’t echo shortcode

don’t render Everwhere when no API key is set

allow Chatbots without instruction to render

Everywhere mode also works when chatbot is present on a page

UI improvements

‘threadId’ not being sent correctly when using and Assistant + Streaming

chatbot shortcode not rendering

Rate limit for Chatbots

render Chatbot as full sized embed

use Chatbots in Chats

faulty Chats resulting in fatal error

Error notice potentially showing for some pages

Chatbot not working when everywhere mode is enabled

Chatbot condition to check if currently on a specific admin page

Chatbot condition to check if page is block editor

SleekMotion not loading in admin when Everywhere mode is enabled

add copy to clipboard button to messages

add modal with possible meta keys when using post meta/query as data source

automatically set turbo models when using assistants

chat will show error if streaming fails

convert post content to Gutenberg blocks when using post content data source

only show GPT-* models in model dropdown

only show turbo models when using Assistants

strip HTML tags from post content when using data post content data source

UI potentially breaking when using external plugins

Bricks content as single data source and query data source

Elementor content as single data source and query data source

Everywhere mode

Oxygen content as single data source and query data source

add link to docs in settings under ‘Streaming’ switch

prevent .js bundle clashing with other JS on the page

ask for confirmation before deleting data rows

better Chatbot UI

don’t automatically fold admin menu bar

minify assets

ordered list didnt have correct color

set default model to gpt-3.5-turbo

show error boundary when chat crashes

Chatbot not using selected model in preview

admin condition to display Chatbots in dashboard

Assistants integration

Logo chatbot button type

possibility to add logo to Chatbot

possibility to stream messages

query as data source for Chatbot

chat breaking after second message

always scroll down to latest message when using chatbot on mobile

better chatbot UX on mobile

use safe-area-inset for chatbot

prevent title generation wrapped in quotes

chabots

hide Prompts CPT from admin

various UI improvements

settings crashing when an invalid API key is still saved

150+ system prompts

prompt builder

add loader

show errors from API calls

don’t remove API key if quota is exceeded

PHP 7.4 incompatibility

minify assets

model not saving

submit button in message box leading to error

initial release